REFRACTORIES 487 and 1,450°C. for magnesite brick. However, no definite standards have been adopted by any official organization for general use. A modified pressure test has been proposed by Nesbitt and Bell in which the specimens are heated to 1,350°C. and immediately subjected to pressure transmitted through a 2-in. steel ball. The depth to which the ball has been pressed into the specimen is considered to be the criterion of the load resistance of the material. The results of tension tests upon refractories in the hot state are not available in the literature nor are data relating to the transverse strength of bricks and tiles. This is to be regretted since in many furnace constructions transverse loads must be considered. Again, it seems very probable that the compression test of firebricks will ultimately be replaced by one involving transverse stress. Thermal Properties.—The thermal qualities of refractories, specific heat, conductivity and expansion are determined according to the established physical methods. It is evident that these properties are of considerable practical importance. The data available, however, on these subjects are quite meager, especially if it is considered that the structure of the manufactured product, irrespective of its chemical nature, is of paramount importance. Furthermore, these properties are subject to change with temperature and comparatively few constants are at hand to illustrate the character of these relations. It is known that the specific heat and thermal conductivity increase with temperature but the fundamental laws governing these changes have not been established. Furthermore, it must be realized that the structure of all these materials is certain to undergo physical changes which affect the thermal qualities. Specific Heat.—This constant has been determined by heating a specimen of known weight in a furnace and when brought to constant temperature allowing it to drop into a calorimeter. The error due to the heat lost in transferring the specimen is usually greater than has been appreciated. The increase in specific heat with temperature appears to represent in general a linear function, though this has not been established beyond doubt and in some cases it is known that the relation is a more complicated one. Thermal Conductivity.—The methods for determining this constant apparently most in favor at the present time are those involving the measurement of heat loss computed from the input of electrical energy utilized in heating coils surrounded by the material under test.1 Comparative tests are frequently made by measuring the temperature drop between known or equal distances of different materials, heated from a source of constant temperature and brought to thermal equilibrium. In general, heat conductivity decreases with the porosity of the materials. Thermal Expansion.—The linear expansion upon heating specimens to known temperatures by immersion in baths maintained at a constant point may be accomplished by direct measurements using a micrometer microscope, by a system of levers, magnifying the expansion so that it may be read from a scale or, through the use of an optical lever, with mirror, scale and telescope. More refined methods, as that of Fizeau, are scarcely ever used in technical work. Probably one of the best known and most satisfactory methods is that of A. W. Gray,2 using bars 30 cm. long, immersed in an oil bath or heated in an air furnace. The length changes are determined by means of two microscopes clamped to an invar 1 R. A. HORNING, Trans. Am. Ceram. Soe., Vol. 19. 2 ,S'c'i. Papers Noa. 332 and 352, Bureau of Standards.